Advertising for online stores: what works and what eats the budget
A store has one advantage over any other business: the money is visible. Not leads, not calls, but actual revenue attached to an actual ad. And one trap: turnover grows while profit does not, because someone measured ROAS instead of margin.
I have worked with stores across categories: children's books, canvas art, homeware, premium devices. What follows repeats in nearly all of them.
Where to start if there is no advertising yet
- Work out what you can pay for a customer. Not what feels acceptable, but what the margin survives after returns and shipping.
- Track purchases, not clicks: pixel, events, order value. If the platform cannot see the amount, it optimises blind.
- Build the product feed. The boring part everyone skips, and the one that delivers the cheapest sales.
- Turn on remarketing before anything else. People who already viewed a product convert several times cheaper.
- Only then cold traffic and creative testing.
In a store with hundreds of items, the algorithm will find what to show and to whom, provided the product data is complete and current. A poor feed with empty fields and stale prices limits delivery more than any creative does.
Where stores lose money most often
| What happens | What it costs | What to do |
|---|---|---|
| Measuring ROAS instead of margin | Turnover grows, profit does not | Count profit per order, not revenue |
| All spend on one bestseller | Growth hits a ceiling | Expand into categories through the feed |
| Returns not accounted for | Real margin below the planned one | Subtract returns from revenue |
| Product page leaves questions open | Traffic arrives, cart stays empty | Size, timing, returns, answered upfront |
| No free-shipping threshold | Order value lower than it could be | Set it just above your current average |
What budget is needed
It depends on your price point, but as a guide: for the algorithm to exit learning you need roughly 800 euros or more per channel per month in Spain and the EU. Less does not mean slower; less often means the campaign never learns at all.
If there is no budget for two channels, do one properly. Split across Meta and Google, the same money usually performs worse than it would in one place.
What actually worked
- Children's books. The store ran on Instagram alone with around ten sales a month. After a proper site, analytics and managed traffic it reached 10 to 12 orders a day. Full case.
- Canvas art. A store built from nothing in two and a half months: Shopify, 150 works, card payments, fiscal receipts, advertising live. Full case.
FAQ
What counts as a good ROAS?
There is no universal number because it depends on your margin. At 30 percent margin a ROAS of 3 is breaking even; at 70 percent the same ROAS of 3 is solid profit. Start from what you can pay per customer, not from someone else's benchmark.
Meta or Google for a store?
Google captures people already searching for your product; Meta creates demand and works well for impulse and visual categories. With budget for one, start where demand already exists.
I am on Shopify, does that change anything?
For the better: the feed, the pixel and most product-page fixes happen without a developer.
Find out where your budget leaks
Free 30-minute audit. You leave with a verdict and three steps, even if we never work together.
Book the audit Message on WhatsApp